A year (or so) as a Data Scientist

A year flies by, I can’t believe it! Someone pinch me? I started this post a few months after my 1 year work anniversary and now I’m practically approaching the second anniversary. I started as a data scientist on a pricing team at Liberty Mutual and I am still with the same team. There’s been a lot of learning and adjusting to do my job effectively, but I think that is true regardless of the company. I also get asked how I’m adjusting to being away from neuroscience research and if the transition was worth the effort.

Let me address this second part first and I’ll circle back to stuff relating to more day-to-day work happenings.

Here’s a fairly typical conversation that happens:

Person I’m Talking With (PTW) - Where do you work?

Me - Liberty Mutual.

PTW - Oh, what do you do there?

Me - I’m a data scientist.

PTW - Insurance companies have data scientists? What kind of data science happens there?

Me - Well there’s analytics on employee engagement (ie HR stuff), agent engagement and satisfaction (eg customer experience), as well as machine learning in claims, legal (eg relating to fraud detection, fees and billing codes), innovation (eg new products). I do predictive modeling on a pricing team.

PTW - Oh wow, I had no idea. How long have you been doing this?

Me - Just about a year. I’m pretty new to the company and the field.

PTW - What did you do before?

Me - I was a postdoc(oral researcher) at the University of Washington researching Alzheimer’s Disease.

PTW - …

I usually laugh at their bewilderment and from there, we talk about why I made a career transition.

With respect to the effort that I put in to transition away from academia, I am truly happy with my decision. I am thankful to my postdoc adviser for being ok with me not knowing what the next step of my career was. She was willing to help me talk through whatever options I was considering and tried to connect me to people who might be helpful in that field. When it was time to let her know what I’d be leaving in 2 months, I was afraid that she would be upset. Even though I knew I would eventually leave academia, when that “some day in the near future” finally had a date, I felt that I had failed the lab and my advisor. I had been there for almost 3 years and didn’t write any grants or publish any papers. It boiled down to feeling I was an unproductive member of the academic community because I spent so much time in lab that my self worth was tangled with the (lack of) work-related achievements. So much guilt and fear. But she wasn’t upset, or if she was, she kept it to herself and we went into planning mode. I left and things were fine. I recently went back to campus to see a grad student defend her thesis and had a chance to talk to her advisor (a committee member of mine). I basically shared the fears I had during and shortly after my postdoc and she reassured me that different paths were ok, that people gravitate to things they’re interested in and good at. Dang, I needed to hear this more often 6+ years ago!

During the first few months at Liberty, I grappled with questions that people asked centered around a theme of wondering why I would leave a noble pursuit to help a business make money. I was literally trying to cure cancer and unravel the mysteries of dementia and this was a tough one to explain. I’m still working on challenging problems, which I like and is great for keeping my brain engaged. It’s also exciting to know that answers I arrive at using models I built are going live soon for products intended to help people when bad things happen. I think knowing that my work impacts people today is powerful.

I’m working on being comfortable describing what I do and I think it’s getting easier as I do more work in this space. Friends, insurance can be....complicated, so pardon me while I start over with my repository of explanations and analogies. I work on pricing models in the commercial insurance space. A big focus for my team was working on a product aimed to help small business owners (you’re busy getting a new venture off the ground; how high is “figure out what kind(s) of insurance I need” on your priority list?). It’s challenging work on a few different levels! Commercial insurance has an initial challenge of not having as much data (ie claims) as personal insurance. The data we do have is very heterogeneous. If I can try to explain it, the ways in which a house or car differs from another house or car are less varied compared to businesses. It's sort of like the variability between Corgis versus the variability between dogs in general. Another challenge is the regulatory one because each state has its own department of insurance (DOI) and you don’t know for sure what variables or data sources, etc, a DOI will have objections to when filing new rates. If a DOI does have an issue, someone on the team needs to put together documentation responding to the issues in the objection. It really reminds me of putting together a rebuttal to reviewers that are reviewing a manuscript (is there a DOI equivalent to a Reviewer #2??). I’m periodically amused at the parallels I run into between insurance and research.

What I am getting great pleasure in my new role is giving presentations. Maximizing the impact of an analysis or model build means sharing the results or providing updates to stakeholders. I am learning about what matters to my audience and it is extra fun when different stakeholders care about different things. I knew working on science communication skills would be beneficial during grad school and postdoc (I was terrified of public speaking) and it continues to be one of the things I’m grateful for being involved in and continue to be involved with.

In short, I’m excited about the direction of my career path and wonder what the next year holds for me!